By Jason Uechi, Director of Engineering and Data Science, YP Mobile Labs
Worry about a technology revolution, and specifically the type of uprising where new technology takes all the jobs, has been around for quite a while. In the late Renaissance, William Lee, the inventor of the first stocking frame knitting machine was denied a patent due to concern directly from Queen Elizabeth the First that his invention would "deprive (knitters) of employment, making them beggars".
Fast forward past the landing at Plymouth Rock, the Industrial Revolution, and the animated gif, and the concern not only remains, but is amplified by worries of the rise of artificial intelligence in the form of benevolent self-driving cars and rolling robot pizza shops. Software is indeed eating the world, and careful analysis shows no job sector is left unaffected by optimization and automation via ever faster computers and cheaper storage.
For media professionals the job was once clear: find the best outlets to reach your customer’s audience locally, then negotiate the best rates for your customers. But now? We've moved from a world focused on specific media and their respective audiences to a world guided by on- and offline behaviors, likes and habits. That’s enabled modern programmatic solutions to create a steady stream of behavioral profiles that can be linked to specific audiences in specific locales.
A confession: I write the code that makes the robots sing. And while it is both challenging and fun to build complex, multi-layered ad-tech systems to ingest billions of points each day, I'm not here to explain how, nor even rationalize why. Instead, I'm here to explain where that data, and all those algorithms, leave room for a modern media planner to lead a successful local campaign that will thrive amongst the clattering servers and the quiet twinkle of pixel beacons.
The Relationship Between Facts:
Unless you're tossing your entire budget inside someone's walled garden, you're probably managing a set of marketing initiatives that span multiple channels and target audiences from a handful of trusted partners. When trying to assess overall performance across these partners and channels the hardest questions, such as, "Is this campaign contributing to more sales?" can get lost amidst varied measurement methods.
Luckily for the modern media planner, this is hard for software, too. It is a decidedly difficult and inherently human task to synthesize insights across different sources and contexts. By piecing together disparate clues, a skilled planner can bring together observations across a complex, multifaceted local media campaign.
Identifying absolute facts - the "ground truth" - is important, whether that is comparing pre-and-post campaign sales figures or tracking your store visit rate - a measure of how your marketing spend is concretely improving customer foot traffic into specific store locations. Identifying these kinds of core facts is key, as they can then serve as your North Star for guiding evaluation across all channels (display, search or social; digital or otherwise). Your very human horizontal view across all data sources separates man from machine.
Turning Machine Learning into Intuition:
The wealth of data that modern digital marketing generates can be overwhelming. Putting that raw data into a meaningful context can be difficult, and much of the practice of data science today is digging deeper into the algorithmic plumbing to ask and answer simple questions.
How can a media planner find answers? The industry trend towards richer reporting and transparency is key. Merely receiving summary information, such as total impressions and total clicks - sufficient for billing but insufficient for deriving any performance insight - is no longer enough.
For example, location-based mobile campaigns can tether raw campaign performance information to places or events. Reviewing and optimizing your campaign performance via an interactive map allows you to view abstract actions through the lens of your own city, through the lens of your own experiences. Data that can be hard for a pattern-matching algorithm to interpret can be completely trivial for someone who can glance at a neighborhood map and instantly say "Well, that's because of traffic off the 405, which I drive on every day."
The modern media planner must find and utilize reporting that combines campaign data and local real-world context, and turn this data into practical insights and optimizations.
Above all, Being Human:
In the end, understanding what makes a local media campaign successful is fundamentally an attempt at understanding human behavior. And despite our best attempt to interpret humans as rational actors, fulfilling our own economic or evolutionary needs, human behavior is often triggered by emotional factors not easily teased into data in machine-learning-friendly form.
In order to thrive, modern media planners must embrace their natural creativity and humanity, alongside digital and programmatic riches. Don't compete with a computer on repetitive tasks, nor with comparing numbers from compatible sources. Instead, let the machines extend efforts at near light-speed scale, but then demand rich, transparent reporting that combines campaign data and intelligent context. Through your analysis, strive to find simple timeless answers, albeit obtained through incredibly complex modern means.
The programmatic revolution may seem like an opaque contraption, a sealed black box of pernicious algorithms and endless data. But for a modern media planner, at the fuzzy edges of the black box is exactly where the warmth of humanity shines. I, for one, welcome the return of the human overlords.